It's not what you might think. I used to assume that the petechiae were the result of an inability of so few platelets to form viable plugs at sites of vascular injury. Red cells would leak out through the holes that were formed after trauma, I thought, thus creating petechiae.
But a review article in this month's New England Journal of Medicine (N Engl J Med 2008;359:1261-70) gives a better explanation.
One of the most important agents to cause one vascular endothelial cell to bind to the next is the transmembrane protein cadherin complex. This acts as a type of adhesive between the endothelial cells. As it turns out, platelets seem to constitutively release substances that bind to specific endothelial receptors and, via complex complex intermediate pathways, stabilise these cadherin complexes.
So far, so good - in health. But when platelet numbers drop below a certain threshold, there aren't enough of them anymore to maintain the cadherin complex. The result of this is that this intracellular barrier is lost, which permits passive extravasation of red blood cells through the gaps. Note that, unlike my theory, this implies that petechiae aren't necessarily formed in places where there has been local trauma - petechiae can form in entirely undamaged endothelium, via the gaps that open up during thrombocytopenia.
In many ways, this better theory explains more clinical findings. For one, petechiae are often widespread and diffuse, a finding that is more parsimoniously explained in the above manner, rather than by assuming that every one of these areas are subjected to minor mechanical damage. Also, aspirin and other anti-platelet drugs, while often associated with an increased risk of bleeding, are not usually associated with petechiae. This is no doubt because these drugs are capable of inhibiting the platelet's ability to form a haemostatic plug, but not their ability to nourish the vascular endothelium.